Lack of correlation between SREBF1 genotype and hyperlipidemia in individuals treated with highly active antiretroviral therapy

Understanding the mechanisms of hyperlipidemia in patients receiving highly active antiretroviral therapy (HAART) could assist in the treatment of current patients and the development of future ‘lipid neutral’ treatments. A pharmacogenetic approach can allow for such mechanistic insight; the correlation of a genetic variant with the degree of hyperlipidemia observed during treatment implies that the gene product may be part of a relevant pathway affected by a drug. Sterol response element binding protein 1c (SREBP-1c, gene SREBF1) is a transcription factor that activates genes of lipid metabolism [1]. Recently, a common variant in the 3’ untranslated region of SREBF1 (SREBF1 3′322C > G) was identified and associated with cholesterol and triglyceride changes in HAART patients [2]. We prospectively evaluated the association of the SREBF1 genotype with elevations of total cholesterol or triglycerides in a large clinical trial that compared the safety and efficacy of HAART regimens containing stavudine, lamivudine, and either lopinavir–ritonavir or nelfinavir for the initial treatment of HIV infection [3]. The pharmacogenetic analysis subgroup (n = 355) was representative of the entire study population (n = 653) according to demographic and lipid characteristics. SREBF1 genotypes were determined by pyrosequencing. Assay details are available on request. Genotypers were blind to the clinical results associated with the DNA samples. Data were analysed using SAS 6.12 (SAS Institute, Cary, NC, USA). The analysis plan was pre-specified. Analysis endpoints were the 48-week maximum change from baseline of total cholesterol and triglycerides. Sub-study subjects dosed with lopinavir–ritonavir and nelfinavir had similar baseline lipids, and a similar maximum change of total cholesterol distributions. The mean maximum change of triglycerides was significantly higher in lopinavir–ritonavir patients. Accordingly, data were analysed by two-way analysis of variance, with drug treatment as a block effect. This analysis had a greater than 80% power to detect a difference of means between two genotypes of 22 mg/dl for total cholesterol or 147 mg/dl for triglycerides. The maximum change of total cholesterol within 48 weeks of HAART treatment was not significantly dependent on either genotype (P = 0.12) or treatment regimen (P = 0.67). The mean changes observed in the six subgroups defined by genotype and treatment (e.g. CC genotype in ritonavir–lopinavir patients; CG genotype in nelfinavir patients) were similar, ranging from 63 ± 36 to 75 ± 40 mg/dl (see Fig. 1). Similarly, the maximum change of triglycerides was not significantly associated with the genotype (P = 0.80). Our results demonstrate that the SREBF1 3'322C > G genotype is not predictive of a risk of hyperlipidemia in the study patients. There was no statistically significant or clinically relevant difference between the three SREBF1 genotypes evaluated in the mean maximum change from the baseline of total cholesterol or triglycerides in patients treated with a HAART regimen containing stavudine, lamivudine and either lopinavir–ritonavir or nelfinavir, or in the combined population.Fig. 1.: Lack of dependence of total cholesterol changes in patients treated with highly active antiretroviral therapy on the SREBF1 genotype and treatment regimen. Box-and-whisker plots show the median, quartiles, and range (to 1.5 × interquartile distance beyond the quartile). Three high outliers (genotypes CC, CC, CG) and one low outlier (genotype CC) had cholesterol changes outside the scale of these plots. TC, Total cholesterol.Several factors may account for the difference between our results and those previously published [2]. First, the previous study was smaller (n = 67) with less power to resolve a difference. It is not uncommon to find that preliminary findings from small studies may not be confirmed by larger studies [4]. Second, we used the maximum change across multiple patient visits, providing a better chance of detecting a relationship, if one exists, than the percentage change between two visits separated by a variable length of time. Third, lipid measurements in our study were made without regard to fasting, which better reflects clinical practice. Fourth, we did not group CC and CG genotypes, as the prevrevious study did. In the absence of knowledge that the GG genotype has a recessive function, such grouping is unusual. We observed slight but consistent heterosis in our study; that is, the maximum lipid changes in CG heterozygotes were higher than those in either homozygous group. It would be of interest to learn whether a similar observation was made in the other study. Fifth, we did not exclude patients on the basis of abnormal baseline lipid levels or body mass index. Finally, our study included patients of greater racial diversity, more reflective of actual patient demographics. Our results provide no definitive evidence as to whether SREBP-1c is mechanistically involved in hyperlipidemia in HAART-treated patients. We studied a variant in the 3’ untranslated region of the gene to which no molecular phenotype has been ascribed. This may be non-functional, or the function may be sufficiently subtle that it does not substantially impact the complex phenotype of hyperlipidemia. Because our study included a limited variety of HAART regimens, we cannot formally exclude the possibility that the SREBF1 genotype is relevant for other drugs. Although no relationship was observed between the SREBF1 genotype and lipid elevations in this study, research to discover predictive factors for the risk of hyperlipidemia in HIV-positive patients should continue to be a priority.